I’m always a bit amused by those who appeal to statistical error, claiming the reduced numbers of >7.0 earthquakes in prior decades/centuries were due partly to the recent numerical increase in seismometers.
It is academia at its most ignorant failure to identify their statistics with the things being measured.
If we accepted that logic, then people in remote regions just didn’t have the resources to perceive >7.0 earthquakes, or perhaps forgot about them. Such an absurd notion disproves their assertions prima facie.
Perhaps in very remote regions where nobody ventured within a 1000 miles or traversed the vicinity, and neither were there geologists to have looked at geologic records, some might have been missed,...but even then that is based upon conjecture as well.
From a layman's point of view much of "science" appears to be nothing more than conjecture. In this particular instance I'll accept the notion that the USGS' recorded and assimilated historical earthquake data is more or less accurate. I can't imagine there would be an agenda at work to prompt them to skew the data. AND, the data shows ups and downs in earthquake magnitude and frequency as far back as they have been able to measure but otherwise remarkably consistent on average.